lokalise-core-workflow-a

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the lokalise2 CLI to facilitate translation workflows, such as uploading files and managing keys. This aligns with standard developer operations for the Lokalise service.
  • [EXTERNAL_DOWNLOADS]: The instructions direct the user to install the official @lokalise/node-api package via the standard npm registry to enable SDK-based interactions.
  • [INDIRECT_PROMPT_INJECTION]: The skill reads content from local translation files (JSON, YAML, etc.) to upload them to the Lokalise API. 1. Ingestion points: readFileSync operations in SKILL.md and implementation-guide.md. 2. Boundary markers: Not used for the translation data. 3. Capability inventory: Bash(lokalise2:*), Write, Edit. 4. Sanitization: None performed on translation values. The content is processed as data for external storage rather than instructions for the agent, presenting minimal risk.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:44 AM
Security Audit — agent-trust-hub — lokalise-core-workflow-a